Principles of Ethical and Fair AI Use AIE Practice Questions (Page 3)

Questions 11–15

1. 11application · medium

   A university is developing an AI system to screen student applications. The system uses historical admission data, which reflects past biases against certain demographic groups. The university wants to ensure the system is ethical and fair. Which approach best aligns with ethical decision-making frameworks?

   • AUse the historical data as-is to maintain consistency with past decisions.
   • BConduct a stakeholder impact assessment and involve diverse stakeholders in the design and review process.
   • CKeep the system's decision-making process confidential to avoid controversy.
   • DDelegate all admission decisions to the AI system without human oversight.
   Reveal Solution Discussion Select an answer first
2. 12foundation · easy

   Which step is typically part of an ethical decision-making framework for AI?

   • AIdentifying potential harms and benefits to stakeholders
   • BEnsuring the model runs on the latest hardware
   • CSelecting the most complex algorithm
   • DMaximizing the number of features used
   Reveal Solution Discussion Select an answer first
4. 13foundation · easy

   Why is ethical AI important in the context of artificial intelligence systems?

   • AIt ensures AI systems are technically flawless and free from all errors
   • BIt makes AI systems fully autonomous without the need for human oversight
   • CIt helps prevent harm to individuals and society while building trust in AI technologies
   • DIt guarantees that AI systems will always make the most profitable decisions
   Reveal Solution Discussion Select an answer first
5. 14expert · hard

   A social media company wants to deploy an AI system that automatically detects and removes hate speech. The system is highly accurate for English content but performs poorly on minority dialects and often removes benign content from those groups. The company faces pressure from advertisers to reduce harmful content and from civil rights groups to protect free expression. Which approach best balances the competing ethical considerations?

   • ADeploy the system globally but include a human review process for content flagged in minority dialects.
   • BDelay deployment until the model is perfect for all dialects.
   • CDeploy the system only in English-speaking countries to avoid the dialect issue.
   • DDeploy the system as-is and rely on user appeals to correct mistakes.
   Reveal Solution Discussion Select an answer first
6. 15expert · hard

   An e-commerce company is developing a recommendation system. The data shows that users from a certain region have significantly lower click-through rates, but the company suspects this is due to a lack of product availability in that region, not a lack of interest. The model trained on this data will likely recommend fewer products to users in that region. The company wants to be fair. What is the most appropriate action?

   • ARemove all regional information from the model to ensure equal treatment.
   • BTrain the model on data from all regions equally, ignoring the regional differences.
   • CUse the model as-is, since it reflects actual user behavior.
   • DInvestigate the root cause and, if confirmed, adjust the model to account for the availability issue, such as by using a different training signal.
   Reveal Solution Discussion Select an answer first